On Sat, 30 Nov 2013 11:54:38 +0100 jdd <jdd@dodin.org> wrote:
... if it can be managed, it's an other word of a full rolling release? nothing to upgrade ever (or in other words, smoth regular upgrade) - I would love it :-)
When update works as expected we are all happy, but what if it fails? You have to have backups, and start (few) hours journey to have system working again. In other words, time to recover from failed update is a problem why people do not like them, and why they avoid using systems that will need such recovery too often, like Factory. IMHO, solution is to redesign Factory installation to provide easy fall back that will not last more then reboot to the latest working system. With current hardware and openSUSE, it means lesser than a minute. The easiest way would be to have 2 installations, one that we run, and another that is initially just copy of first, but will be updated. After update, reboot and you are in a brand new system. If it works you keep it, if not back to working and report a problem. This would mean that we have only half of disk space, but with basic RAID we waste space at the same rate and we have no protection from failed updates. BTW, with two systems in place we don't have to update grub. Install it once and after verification that it works, don't touch it. Kernel and initrd can be 2 times 2 symlinks to partitions and real kernels and initrds. Once basic system in Factory is stable enough, branch it to release candidate, and create stable release for all those that need stability.
jdd
-- Regards, Rajko. -- To unsubscribe, e-mail: opensuse-project+unsubscribe@opensuse.org To contact the owner, email: opensuse-project+owner@opensuse.org